Nvidia's new 34B-parameter vision-language-action model, Alpamayo 2 Super, is now commercially available on Hugging Face under OpenMDW-1.1 license, accelerating self-driving tech.
Nvidia has launched Alpamayo 2 Super, a 34 billion-parameter vision-language-action model for Level 4 autonomous driving.
The model is commercially available on Hugging Face under the permissive OpenMDW-1.1 license, enabling widespread adoption and development.
Nvidia has released Alpamayo 2 Super, a 34 billion-parameter vision-language-action model, for commercial use in Level 4 autonomous driving applications. Available on Hugging Face since August 4 under the OpenMDW-1.1 license, this advanced AI is set to power robotaxis and a wide range of autonomous vehicles globally. The model has already demonstrated superior performance, ranking first on Wayve’s LingoQA benchmark for visual question answering.
This release underscores Nvidia's strategic push to democratize sophisticated AI tools, making high-performance autonomous driving capabilities more accessible to developers and automakers. Alpamayo 2 Super is designed to process complex multi-camera video and navigation inputs, generating precise trajectories and reasoning traces essential for safe and reliable self-driving operations. It forms a crucial part of Nvidia’s broader Alpamayo stack, which supports the entire development lifecycle from data labeling to in-vehicle deployment.
Nvidia’s Open Driving Model Could Reshape the Autonomous-Vehicle Stack. Nvidia’s release of Alpamayo 2 Super is strategically significant because it opens a powerful autonomous-driving foundation model to developers, automakers and mobility startups. The 34-billion-parameter system processes surround-view video, motion history and language inputs to generate trajectories, reasoning traces and scene understanding. Its weights are available on Hugging Face under the permissive OpenMDW-1.1 licence, including commercial use and redistribution rights. However, open access should not be confused with road-ready Level 4 autonomy. Nvidia’s benchmark results—including a leading LingoQA score—are promising but remain development metrics, not proof of safe deployment across unpredictable real-world environments. In our view, Nvidia is positioning itself as the foundational infrastructure provider for autonomous mobility. Its success will depend on validation, safety assurance and whether developers can convert this large model into efficient, dependable in-vehicle systems.
The Strategic Impact on the Autonomous Ecosystem
Model Scale and Performance: Alpamayo 2 Super is a 34 billion-parameter model, significantly outperforming competitors like Qwen2.5-VL 72B, Gemini 2.5 Pro, and GPT-4o in Nvidia's Lingo-Judge tests. Its top ranking on Wayve’s LingoQA benchmark, covering 28,000 scenarios and 419,000 annotations, highlights its robustness in visual question answering for autonomous driving.
Open-Source Accessibility: The model's availability under the OpenMDW-1.1 license from the Linux Foundation allows for extensive use and redistribution, fostering broader innovation in the autonomous vehicle space. This permissive licensing structure could accelerate adoption and collaborative development across the industry.
Industry Integration: Alpamayo 2 Super seamlessly integrates with Nvidia’s DRIVE Hyperion platform, which is already a preferred choice for major automakers and robotaxi operators. This includes global players like Mercedes-Benz, Hyundai Motor Group, General Motors, Uber, Toyota, BYD, Geely, Volvo Cars, and JLR, indicating significant market traction and potential for rapid deployment.
I think this move by Nvidia, making such a powerful model openly available, is a clear signal of their intent to dominate the foundational AI layer for autonomous systems, impacting everything from urban robotaxis to next-generation logistics. The industry should watch closely how this open approach accelerates innovation and competition, especially among Southeast Asian and Indian startups focused on last-mile delivery and smart city solutions.
Frequently asked questions
What is Nvidia Alpamayo 2 Super?
Nvidia Alpamayo 2 Super is a 34 billion-parameter vision-language-action (VLA) model specifically designed for Level 4 autonomous driving. It enables self-driving systems to understand and interact with their environment more comprehensively.
What level of autonomous driving does Alpamayo 2 Super support?
Nvidia Alpamayo 2 Super is developed to support Level 4 autonomous driving, which means the vehicle can handle most driving situations independently, though a human driver may still be required in certain conditions.
Where is Alpamayo 2 Super commercially available?
Alpamayo 2 Super is commercially available on Hugging Face, a popular platform for machine learning models and datasets. This allows for easy access and integration for developers.
Under what license is Alpamayo 2 Super released?
The model is released under the permissive OpenMDW-1.1 license. This license facilitates widespread adoption and encourages development by providing flexibility for commercial and research use.
What does 'vision-language-action model' mean?
A vision-language-action (VLA) model like Alpamayo 2 Super can process visual information, understand natural language instructions, and generate appropriate actions. This comprehensive capability is crucial for complex tasks like autonomous driving.
How many parameters does Alpamayo 2 Super have?
Nvidia's Alpamayo 2 Super boasts 34 billion parameters. This large parameter count contributes to its advanced capabilities in understanding and interacting with complex autonomous driving scenarios.
